TakaGoto

quiz

"Test your understanding of a RAG concept"

TakaGoto 17 4 Updated 5mo ago
GitHub

Install

npx skillscat add takagoto/rag-learning-academy/quiz

Install via the SkillsCat registry.

SKILL.md

Quiz: Test Your RAG Knowledge

Scope: This skill tests your understanding with questions and scoring. For guided teaching of new concepts, use /lesson.

Generate and administer a quiz on a specific RAG topic to help learners assess their understanding. Quizzes reinforce learning and surface gaps.

Step 1: Determine the Topic

  • If the user specifies a topic (e.g., /quiz chunking), use that topic.
  • If no topic is given, read progress/module-tracker.md to find the most recently completed lesson and quiz on that.
  • If no progress exists, suggest running /start or /lesson first.

Step 2: Generate Questions

Create 5-10 questions that mix the following types:

Conceptual Questions (2-3)

Test understanding of "why" and "how" things work. Example: "Why might fixed-size chunking lose important context at chunk boundaries?"

Code Reading Questions (2-3)

Show a code snippet and ask what it does, what is wrong with it, or what the output would be. Example: Show a retrieval function and ask why it might return irrelevant results.

Scenario-Based Questions (2-3)

Present a real-world situation and ask the learner to choose the best approach. Example: "You have a corpus of legal contracts averaging 50 pages each. Which chunking strategy would you start with and why?"

True/False with Justification (1-2)

A statement the learner must evaluate and explain. Example: "Larger chunk sizes always improve retrieval quality. True or false? Explain."

Step 3: Administer the Quiz

Present questions one at a time. Wait for the learner's answer before moving to the next question. Do not reveal the correct answer until the learner has responded.

For each question:

  1. Present the question clearly
  2. Wait for the learner's response
  3. Evaluate the response — is it correct, partially correct, or incorrect?
  4. Provide the correct answer with a clear explanation
  5. If incorrect, explain the misconception and link it back to the relevant concept

Step 4: Score and Review

After all questions are answered, present a summary:

Quiz Results: [Topic]
Score: [X]/[Total] ([percentage]%)
---
Strengths: [areas where the learner did well]
Gaps: [areas that need review]
Recommendation: [what to study next]

Step 5: Track Results

Save the quiz results to progress/quiz-results.md with the date, topic, score, and any identified gaps. Append to the file if it already exists.

Step 6: Suggest Next Steps

Based on the score, suggest 2-3 relevant next steps using slash commands:

  • 90-100%: Congratulate them.
    • /challenge — put your mastery to the test with a real-world challenge
    • /build — apply your knowledge by building a hands-on component
    • /lesson — continue to the next lesson in your track
  • 70-89%: Good progress. Recommend reviewing the specific gaps.
    • /build — reinforce your understanding by building what you just learned
    • /lesson — continue to the next lesson or revisit a weak area
    • /explain — get a deeper explanation of any concepts that felt shaky
  • Below 70%: Encourage them.
    • /lesson — revisit the lesson to strengthen your foundations
    • /explain — get a deep-dive explanation on the weak areas
    • /glossary — review key terms you may have missed

Guidelines

  • Questions should be practical and relevant, not trivia
  • Accept reasonable alternative answers — RAG is a field with multiple valid approaches
  • When answers are wrong, be straightforward: explain what's actually going on without sugarcoating. "Not quite — here's the thing" beats "Great try! You're on the right track!"
  • Vary difficulty within the quiz (start easier, build up)